Abstract
This paper presents an energy-ef ficient processing platform for wearable sensor nodes, designed to support diverse biological signals and algorithms. The platform features a 0.5 V–1.0 V 16-bit microcontroller, SRAM, and accelerators for biomedical signal processing. Voltage scaling and block-level power gating allow opt imizing energy ef ficiency under applica- tions of varying complexity. Programmable accelerators support numerous usage scenarios and perform signal processing tasks at 133 to 215 lower energy than the general-purpose CPU. When running complete EEG and EKG applications using both CPU and accelerators, the platform achieves 10.2 and 11.5 energy reduction respectively compared to CPU-only implementations.